FP-MaxwellPretzlav

From CS294-10 Visualization Fa08

Project Proposal

Group Member

Maxwell Pretzlav

Description

As stated in my Assignment 3 proposal, Last.fm has a vast amount of data on my, and many other people's, individual music listening history. In assignment 3, Nick Kong and I made a simple stacked-graph visualization for two sets of a single user's listening history over the last year.

I intend to expand and enhance this visualization. My main interest is allowing the user to select different visualization methods for the same data, as different views will afford examining different aspects of the data. Since Lee Byron's Streamgraph there have been a number of attempts to implement streamgraphs of last.fm data. While streamgraphs are interesting and useful [1], interactive visualizations needn't be limited to a single visualization method. For instance, stacked graphs are not terribly useful for comparing two different measurements of values (i.e. artist playcounts) at the same time-point, since that is a visual judgment of length, rather than position.

Due to its hierarchical nature, this data affords some unique visualizations that I haven't seen much discussion of. While our first prototype had a simple one-level drill-down from tags to artists, there are many layers that could be visualized: tag → artist → album → track. Additionally, while playcounts are interesting, they can also be misleading as they don't account for the actual length of each track. The MusicBrainz API allows access to their database of track lengths, which I intend to use to calculate actual time listened to.

There's a lot of room for using interactivity to display this data in multiple ways to find new and unusual patterns. Last.fm's data affords visualizations comparing the same time period between multiple users, something I imagine many users would be interested in seeing.